Julian Hall
This closes #1962
This will close #2423
One for you @galabovaa, in due course
Closed by #2428
I guess this is related to #2394 @galabovaa
From #2404 Dear HIGHS; Previously with HIGHS 1.10.0, it was possible to interface directly in Fortran integer-feasible solution callbacks and then to retrieve the integer-feasible-solution's mip_solution() 1D-array via the HighsCallbackDataOut...
The short answer is that 56% (605612/108000) of your coefficient matrix entries are nonzero, so presolve is unlikely to reduce the size of the problem to be solve. All that...
The culprit (@fwesselm) is probing. I've added another bit to the `presolve_rule_off` option so that probing can be prevented. This leads to presolve taking 30s, and the MIP is solved...
Very interesting, I'd not considered that consequence. Indeed, for your purposes reduction_limit=0 is the way to go
Clearly HiGHS gets too excited about probing, and it will be good to find out why. In particular to find out why the memory use snowballs. CBC may gain by...